AI Voice Disclosure

Purpose, Scope, and Relationship to Other Terms

This AI Voice Disclosure applies to Ask Bennett website voice assistants, telephone voice assistants, chat experiences that include generated speech, and related recordings, transcripts, summaries, routing, appointment, transfer, and support features.

It explains the service to customers, website visitors, callers, and other people who interact with Bennett. It also establishes minimum disclosure and configuration standards for customers deploying Bennett. Contractual rights, payment obligations, liability limitations, indemnification, suspension rights, and dispute terms are governed by the Terms of Service and Service Agreement.

If this disclosure conflicts with mandatory law, mandatory law controls. Customers must use any additional notice, consent, opt-out, recording announcement, or wording required for their location, audience, industry, telephone campaign, or use case.

Bennett Is an AI Assistant, Not a Human

Bennett is software that uses artificial intelligence, automated speech recognition, language processing, and generated voice to respond to questions and perform configured tasks. Bennett is not an employee, licensed professional, human operator, emergency dispatcher, or independent decision-maker.

A natural-sounding voice, conversational timing, name, avatar, greeting, or first-person wording does not make Bennett human. Customers must not describe Bennett as a real person, conceal Bennett’s AI nature, instruct Bennett to deny being AI, or use presentation choices that are reasonably likely to mislead a person into believing a human is speaking.

Required Opening Disclosure

Each customer must provide a clear AI identity disclosure at the beginning of a voice interaction. When recording or transcription is enabled, the opening must also disclose those practices and obtain any consent required by applicable law before recording begins or before substantive personal information is collected.

The recommended wording is a baseline, not a guarantee of compliance in every jurisdiction. Customers must adapt it to their actual configuration, recording settings, transfer options, business name, legal requirements, audience, and industry. A customer must not state that recording, human transfer, or another feature is available unless it is actually enabled.

Disclosure Timing, Visibility, and Accessibility

For telephone interactions, the AI identity notice should be delivered at or near the start of the call, before substantive conversation. For website voice interactions, the interface should visibly identify Bennett as AI before activation, and the spoken opening should repeat that identity when the voice session begins.

Disclosures must be prominent, understandable, and appropriate to the medium. They must not be hidden in a footer, privacy policy, tooltip, rapidly spoken message, low-contrast text, or wording that a reasonable person is unlikely to notice or understand.

When an interaction is directed to or likely to include children, older adults, people with disabilities, or other potentially vulnerable users, the customer must use language and presentation appropriate to the audience and must obtain separate written approval from Ask Bennett before any child-directed deployment.

AI-Generated and Synthetic Voice

Bennett’s speech may be generated by text-to-speech or other synthetic-voice technology. The voice may include natural cadence, pauses, tone, accent, pronunciation, and conversational cues, but it is produced by software.

Generated speech may mispronounce names, numbers, addresses, technical terms, dates, prices, or unfamiliar words. Background noise, connection quality, accents, speech differences, and interruptions may also affect what Bennett hears and says.

Customers must not use a synthetic voice to falsely imply endorsement, authority, affiliation, or identity. A voice should not be represented as belonging to a real person unless the customer has documented rights, authorization, and all legally required consents, and Ask Bennett has approved that use in writing.

Approved Website and Business Source Content

Bennett is configured to answer from customer-approved sources. The primary business-specific source is the customer’s website. A customer may also expressly provide or approve additional materials, instructions, schedules, forms, knowledge entries, integrations, or data for use where supported.

Bennett does not independently verify that source content is complete, accurate, lawful, current, noninfringing, or appropriate. The customer owns and controls its business content and remains responsible for every statement, price, policy, availability representation, instruction, product description, disclaimer, and other item supplied or approved for Bennett.

Ask Bennett may use technical retrieval, indexing, prompting, search, and model services to locate and present approved information. These processes do not guarantee that every response will reproduce the source exactly or select the best available source passage.

Information Bennett May Not Know

If information is absent from the approved website or other approved source, inaccessible to the system, blocked by technical settings, recently changed, contradictory, poorly written, or outside the configured scope, Bennett may not know it or may be unable to answer correctly.

Bennett does not automatically know private business practices, internal decisions, real-time inventory, unwritten exceptions, employee knowledge, unpublished pricing, changing availability, account-specific facts, or information stored in systems that have not been properly integrated and authorized.

AI Errors, Inaccuracies, and Unexpected Output

Artificial intelligence can make mistakes. Bennett may misunderstand a question, mishear speech, combine unrelated information, omit a condition, rely on outdated content, infer something not stated, provide an incomplete response, translate incorrectly, repeat information inaccurately, or generate a response that is unexpected, inconsistent, or false.

These issues may be described as AI errors, hallucinations, model errors, retrieval errors, speech-recognition errors, transcription errors, configuration errors, translation errors, or output errors. No AI system can be guaranteed to be error-free, uninterrupted, complete, or correct in every context.

Customers must not configure or market Bennett as infallible, guaranteed accurate, a substitute for a human professional, or capable of making binding promises unless a qualified human independently approves the promise through a controlled workflow.

Human Verification and Oversight

Important information should be independently verified with the business or an appropriately qualified human. This includes prices, promotions, availability, eligibility, contractual terms, legal rights, health or safety instructions, financial consequences, deadlines, account status, refunds, cancellations, appointments, and any statement that could materially affect a person.

Customers must test Bennett before launch, review representative conversations, correct source content, monitor recurring errors, maintain escalation procedures, and retrain or reconfigure the experience when business information changes.

A person’s failure to verify an AI response does not convert Bennett’s output into a warranty, binding professional opinion, guaranteed business commitment, or verified fact.

Call Recording, Transcription, and Consent

Depending on the customer’s configuration, a voice interaction may be recorded, transcribed, summarized, analyzed, or stored. Recording and interception laws differ by country, state, province, territory, and context. Some laws require consent from one participant; others may require consent from every participant or a specific announcement.

The customer is responsible for determining the applicable rule, providing the required notice, obtaining and documenting valid consent, respecting a refusal, and configuring the service accordingly. A generic statement in this page does not replace an in-call or on-screen notice when the law requires notice at collection.

Where a person does not consent and consent is legally required, the customer must provide a compliant alternative, disable the recording or transcription where technically supported, transfer to a human where available, or end the interaction without collecting additional information.

Voice, Transcript, and Conversation Data

Voice audio, speech segments, transcripts, chat text, prompts, responses, telephone metadata, timestamps, summaries, routing information, appointment details, and technical logs may be processed to deliver, secure, troubleshoot, support, and improve the configured service as described in the Privacy Policy and Data Processing Addendum.

Customers should collect only information reasonably necessary for the stated purpose. They must not invite passwords, full payment-card numbers, private keys, authentication codes, government identification numbers, protected health information, biometric identifiers, or other highly sensitive information unless the use has been separately approved, legally reviewed, and configured with appropriate safeguards.

End users should avoid sharing unnecessary sensitive information with Bennett. Payment information should be entered only through an approved secure payment interface, not spoken into a general voice conversation unless a separately approved compliant payment workflow is used.

Biometric Voiceprints, Identification, and Emotion Inference

Ask Bennett is not intended by default to create biometric voiceprints, identify a person by voice, authenticate identity through voice, infer protected traits, or determine emotion, mental state, truthfulness, or intent from vocal characteristics.

Customers may not use Bennett for biometric identification, speaker recognition, emotion recognition, lie detection, sensitive-trait inference, or voice-based profiling without Ask Bennett’s prior written approval, a documented lawful basis, required notices and consents, retention and deletion rules, security controls, and specialized legal review.

Ordinary processing of audio to recognize words or generate speech does not mean that no biometric law can apply. The legal classification depends on the technology, purpose, data, and jurisdiction, and customers remain responsible for their deployment.

Voice Cloning, Identity Rights, and Impersonation

Customers must not clone, imitate, reproduce, or deploy the voice or persona of an employee, executive, celebrity, public official, family member, customer, competitor, or other person without documented authorization and all rights required by contract, publicity, privacy, biometric, intellectual-property, employment, and consumer-protection law.

Bennett may not be used to falsely pose as a government agency, business, professional, officer, employee, relative, or other person, or to create a false impression of endorsement, sponsorship, affiliation, urgency, or authority.

Ask Bennett may require proof of authorization, reject a voice, remove content, preserve evidence, suspend a deployment, or report suspected fraud, exploitation, or unlawful impersonation.

Inbound Website and Telephone Interactions

For inbound interactions initiated by a visitor or caller, the customer must still disclose that Bennett is AI and disclose recording or transcription when applicable. A person choosing to click a microphone, open a voice widget, or call a published number does not automatically waive disclosure, consent, privacy, accessibility, or consumer-protection requirements.

Bennett may answer common questions, route inquiries, collect limited information, schedule appointments, transfer calls, or take other configured actions. The customer must ensure each action is accurately described, properly connected, and subject to human review where necessary.

When Bennett cannot answer or the user disputes the answer, the interaction should not pressure the person to accept the output as correct. The customer should provide a practical way to reach the business, submit a correction, or end the interaction.

Outbound AI-Generated Voice Calls

Outbound calling may be restricted or unavailable unless separately enabled and approved. In the United States, the Federal Communications Commission has confirmed that AI-generated voices are treated as artificial or prerecorded voices under the Telephone Consumer Protection Act. Consent, identification, opt-out, do-not-call, calling-hour, recordkeeping, and other rules may apply.

Other countries and subnational jurisdictions impose their own telemarketing, electronic-communications, direct-marketing, consent, suppression, and caller-identification requirements. A customer must not initiate an outbound AI voice call merely because it possesses a telephone number or has a general business relationship.

Before any outbound deployment, the customer must obtain written approval from Ask Bennett, document the lawful basis and required consent, maintain suppression records, honor opt-outs promptly, use accurate caller identification, and comply with the Acceptable Use Policy.

Language, Accent, Translation, and Cultural Context

Bennett may support multiple languages and accents, but availability and performance vary. Speech recognition, translation, pronunciation, tone, idioms, names, dialects, and culturally specific meanings may be inaccurate or misunderstood.

A language option does not certify legal, professional, or native-level translation. Customers must have important multilingual content reviewed by a qualified person and must not rely on Bennett as the sole translator for contracts, legal notices, medical instructions, safety warnings, regulated disclosures, or other material information.

Customers should make the AI identity, recording notice, privacy information, and opt-out path available in the language used for the interaction whenever required or reasonably necessary for understanding.

Accessibility and Equal Access

Customers should provide accessible ways to understand that Bennett is AI and to access the same material information without voice. Website deployments should support a typed or readable alternative where available, and telephone deployments should offer a reasonable alternative when a disability, speech difference, hearing difference, language barrier, or technology limitation prevents effective use.

Customers may not use Bennett to exclude, discourage, disadvantage, or discriminate against a person because of disability or another protected characteristic. Accessibility obligations vary by jurisdiction and industry and remain the customer’s responsibility.

Human Transfer, Message Taking, and Ending the Interaction

Where the customer enables a human-transfer, callback, message, ticket, email, or other escalation path, Bennett may offer that option. Availability depends on configuration, staffing, business hours, telephone systems, third-party services, and the customer’s response process.

Customers must not promise an immediate transfer, callback time, appointment, resolution, refund, or human review unless that promise is supported by the actual workflow. If no human path is available, the disclosure and interface should not imply otherwise.

A user may stop speaking, close the widget, hang up, or request that the interaction end. Customers must not configure Bennett to obstruct a lawful opt-out, repeatedly pressure a person to continue, or misrepresent that continued interaction is required.

Professional Advice and High-Risk Uses

Bennett provides general business information and configured assistance. Bennett is not a substitute for a lawyer, doctor, pharmacist, therapist, accountant, financial adviser, insurance professional, licensed contractor, government official, or other qualified professional.

Bennett may not be used to make or materially determine eligibility, ranking, pricing, approval, denial, termination, diagnosis, treatment, triage, credit, insurance, employment, housing, education, public-benefit, legal, or other high-impact decisions about a person unless separately approved in writing and lawfully configured with appropriate human oversight.

Customers in regulated industries must obtain their own legal and professional review. Ask Bennett may impose additional terms, safeguards, feature restrictions, testing, documentation, or a complete prohibition for a proposed use.

No Emergency Service

Bennett is not an emergency service, crisis line, dispatch center, safety monitor, or substitute for local emergency professionals. Bennett may fail to recognize an emergency, misunderstand urgency, provide an incorrect location, or be unavailable because of technical interruption.

Customers must not deploy Bennett as the sole channel for emergency reporting, medical triage, suicide or crisis intervention, police, fire, rescue, security response, or time-critical safety instructions.

Reputation, Business Decisions, and Reliance on Output

Because Bennett answers from customer-controlled content and can produce AI errors, the customer is responsible for reviewing how Bennett represents the business. Ask Bennett does not guarantee that an output will improve reputation, prevent a complaint, preserve a customer relationship, create a sale, avoid a negative review, or satisfy every visitor or caller.

To the fullest extent permitted by law and subject to the governing agreement, Ask Bennett is not responsible for reputational harm, adverse reviews, lost sales, lost opportunities, customer dissatisfaction, misunderstandings, decisions made in reliance on output, or claims arising from inaccurate, incomplete, unlawful, outdated, or missing customer content.

The detailed warranty disclaimers, liability limitations, indemnification obligations, mandatory-law carveouts, and exclusive remedies are stated in the Terms of Service and Service Agreement and are incorporated into this disclosure by reference.

Customer Implementation and Ongoing Duties

Each customer is responsible for lawful, accurate, and transparent deployment. At a minimum, the customer must:

• identify the business accurately and disclose that Bennett is an AI assistant;
• provide recording and transcription notices and obtain consent where required;
• publish and maintain accurate website and approved source content;
• test common questions, edge cases, transfers, appointments, numbers, prices, and legal notices before launch;
• monitor output, correct recurring issues, and update the configuration when business information changes;
• provide appropriate human oversight and escalation for material or disputed information;
• avoid restricted, deceptive, discriminatory, unsafe, and high-risk uses;
• protect account credentials and limit administrative access;
• honor privacy, access, correction, deletion, recording, opt-out, and complaint rights; and
• retain evidence of consent, authorization, testing, and compliance when legally required.

Ask Bennett may provide templates, suggested language, settings, or implementation guidance, but the customer remains responsible for determining whether those materials satisfy the law and facts of its deployment.

Jurisdictional Transparency Standards

Ask Bennett uses a clear-disclosure baseline across all supported markets. The following summary is not exhaustive and does not replace local legal advice.

Where multiple laws apply, the customer must follow the requirement that provides the applicable protection. Ask Bennett may require a stricter uniform disclosure to simplify cross-border deployment.

Review, Correction, Suspension, and Enforcement

Ask Bennett may review a deployment, opening script, voice, prompt, source content, recording configuration, calling practice, or complaint when reasonably necessary to protect users, comply with law, enforce the Agreement, or satisfy a provider requirement.

We may require a corrected disclosure, additional consent, removal of a voice, revised content, feature restriction, human-oversight control, temporary suspension, or termination. Urgent action may be taken without advance notice when a use appears deceptive, unlawful, unsafe, exploitative, fraudulent, privacy-invasive, or likely to cause material harm.

A customer-caused restriction, suspension, or termination does not create a refund or credit except where mandatory law requires otherwise. The customer remains responsible for obligations and amounts accrued under the Agreement.

Changes to This Disclosure

We may update this disclosure to reflect changes in law, regulatory guidance, product functionality, recording practices, subprocessors, languages, safety controls, risks, or business operations. The revised page will display a new effective date.

Customers must review updates and modify their live scripts, notices, settings, source content, and operating procedures when necessary. Continued use is governed by the update provisions in the Terms of Service and Service Agreement.

AI Voice, Privacy, and Disclosure Contact

Questions about Bennett’s AI identity, voice generation, recording, transcription, source content, inaccurate output, accessibility, or a planned deployment may be sent to: